BACKGROUND: Rapid improvements in computer technology allow us to consider the use of computer-assisted learning (CAL) for teaching technical skills in surgical training. The objective of this study was to compare in a prospective, randomized fashion, CAL with a lecture and feedback seminar (LFS) for the purpose of teaching a basic surgical skill. METHODS: Freshman medical students were randomly assigned to spend 1 hour in either a CAL or LFS session. Both sessions were designed to teach them to tie a two-handed square knot. Students in both groups were given knot tying boards and those in the CAL group were asked to interact with the CAL program. Students in the LFS group were given a slide presentation and were given individualized feedback as they practiced this skill. At the end of the session the students were videotaped tying two complete knots. The tapes were independently analyzed, in a blinded fashion, by three surgeons. The total time for the task was recorded, the knots were evaluated for squareness, and each subject was scored for the quality of performance. RESULTS: Data from 82 subjects were available for the final analysis. Comparison of the two groups demonstrated no significant difference between the proportion of subjects who were able to tie a square knot. There was no difference between the average time required to perform the task. The CAL group had significantly lower quality of performance (t = 5.37, P <0.0001). CONCLUSIONS: CAL and LFS were equally effective in conveying the cognitive information associated with this skill. However, the significantly lower performance score demonstrates that the students in the CAL group did not attain a proficiency in this skill equal to the students in the LFS group. Comments by the students suggest that the lack of feedback in this model of CAL was the significant difference between these two educational methods.